DeepSWE 1.1
A coding benchmark for software engineering agents and an updated version of the DeepSWE series. It primarily assesses completion of engineering tasks in multi-repository, multi-dependency settings.
Timeline (last 90 days)
Wonder Full scored 57.5 on the SWE-Marathon benchmark and reached 79.0 on DeepSWE 1.1.
On DeepSWE 1.1, Muse Spark Open scores 60 and Kimi K3 scores 68.
In the DeepSWE 1.1 benchmark, Muse Wonder Full achieved 79.0%, exceeding Claude Opus 5.5's 74.2% and GPT-6 Astra's 74.1%.
On DeepSWE 1.1, Muse Wonder beats Claude Opus 5.5 78.8 to 74.2, leads 10.0 to 4.2 on the Harvey legal agent benchmark LAB-AA, and leads 53.1 to 46.3 on SWE-Marathon.
On the DeepSWE 1.1 coding benchmark, Mistral Large 4 Preview scored 62, ranking second, behind only Kimi K3 paired with Kimi Code CLI at 68.
Google reports that Gemini 4 Argon achieved a score of 77.9% on the DeepSWE 1.1 software engineering benchmark (as self-reported by the vendor)
All Facts (4)
Wonder Full scored 57.5 on the SWE-Marathon benchmark and reached 79.0 on DeepSWE 1.1.
50%UnverifiedOn DeepSWE 1.1, Muse Spark Open scores 60 and Kimi K3 scores 68.
50%UnverifiedOn DeepSWE 1.1, Muse Wonder beats Claude Opus 5.5 78.8 to 74.2, leads 10.0 to 4.2 on the Harvey legal agent benchmark LAB-AA, and leads 53.1 to 46.3 on SWE-Marathon.
50%UnverifiedGoogle reports that Gemini 4 Argon achieved a score of 77.9% on the DeepSWE 1.1 software engineering benchmark (as self-reported by the vendor)
50%